Knowledge representation

What to Represent?
Let us first consider what kinds of knowledge might
need to be represented in AI systems:
Objects
-- Facts about objects in our world domain. e.g. Guitars
have strings, trumpets are brass instruments.
Events
-- Actions that occur in our world. e.g. Steve Vai played
the guitar in Frank Zappa's Band.
Dr. AMIT KUMAR, Dept of CSE, JUET, Guna

Performance
-- Abehavior like playing the guitar involves knowledge
about how to do things.
Meta-knowledge
-- knowledge about what we know.

Thus in solving problems in AI we must represent
knowledge and there are two entities to deal with:
Facts
-- truths about the real world and what we represent.
This can be regarded as the knowledge level
Representation of the facts
which we manipulate. This can be regarded as the
symbol level since we usually define the representation in
terms of symbols that can be manipulated by programs.

We can structure these entities at two levels
the knowledgelevel
-- at which facts are described
the symbol level
-- at which representations of objects are defined in terms
of symbols that can be manipulated in programs.

• Mapping between Facts and
Representation

Representation of Facts

English or natural language is an obvious way of
representing and handling facts.
Logic enables us to consider the following fact:
spot is a dog as dog(spot) We could then infer that all
dogs have tails with:
Vx : dog(x) hasatail(x) We can then deduce:
hasatail(Spot)
• Facts may have many to many mapping to
representation
•Ex: All doges have tails and Every dog has a tail

We have briefly mentioned where knowledge is used in AI
systems. Let us consider a little further to what applications and
how knowledge may be used.
Learning -- acquiring knowledge. This is more than simply
adding new facts to a knowledge base.
Classification: New data may have to be classified prior to
storage for easy retrieval, etc.
Reasoning -- Infer facts from existing data.

If a system only knows:
Davis is a Jazz Musician.
All Jazz Musicians can play their instruments well.
If things like Is Davis a Jazz Musician? or Can Jazz
Musicians play their instruments well? are asked
then the answer is readily obtained from the data
structures and procedures.
However a question like Can Davis play his
instrument well? requires reasoning.

The following properties should be possessed by a knowledge
representation system.
Representational Adequacy
-- the ability to represent the required knowledge;
Inferential Adequacy
- the ability to manipulate the knowledge represented to produce
new knowledge corresponding to that inferred from the original;
Inferential Efficiency
- the ability to direct the inferential mechanisms into the most
productive directions by storing appropriate guides;
Acquisitional Efficiency
- the ability to acquire new knowledge using automatic methods
wherever possible rather than reliance on human intervention.
To date no single system optimises all of the above

Simple way to store facts.
Each fact about a set of objects is set out systematically in
columns.
Little opportunityfor inference.
Knowledge basis for inference engines.

•It is not possible to answer simple question like “ who is the
heaviest player?
•Require procedure to infer the knowledge

Advantages
1. Databases are well suited to efficiently representing and
processing large amounts of data (and derivation from a
database is virtually independent of its size).
2. We can build on traditional database systems to process more
complex and more powerful representational devices (e.g.
frames).
Disadvantages
1. Only simple aspects of the problem domain can be
accommodated.
2. We can represent entities, and relationships between entities,
but not much more.
3. Reasoning is very simple – basically the only reasoning
possible is simple lookup, and we usually need more
sophisticated processing than that.

• Relational knowledge is made up of objects consisting of
> attributes
> corresponding associated values.
• The representation system should be augmented with inference
mechanisms that can operate on the structure of the
representation.
• To do this, the structure of representation system must be
designed to support the inference mechanism
• One useful dorm of inference is Property inheritance
• elements inherit attributes and values from more general
classes in which they are included
• Object must be organized into of classes and classes must be
arranged in a generalized hierarchy

Lines: represent attributes
Boxed nodes -- objects and values
of attributes of objects.
Values can be objects with
attributes and so on.
Arrows -- point from object to its
value.
•This structure is known as a slot
and filler structure, semantic
network or a collection of
frames.
•Inference is provided by the two
attributes of property inheritance
•Isa : class inclusion
•Instance : class membership

Example
• team(Pee-wee-Reese) = Brooklyn dodgers
• Batting-average(Three-Finger-Brown) = .106
• height(Pee-wee-Reese) = 6-1
• Bats(Three-Finger-Brown) = Right

Ex: All dogs have tails
V(X) : dog(x) -> hasatail(x)
Ex: 1. “Marcus is a man”
2. “All men are mortal”
Implies:
. 3. “Marcus is mortal”
• Facts represented in a logical form, which facilitates reasoning.
• An inference engine is required.
Advantages:
Aset of strict rules.
Can be used to derive more facts.
Truths of new statements can be verified.
Guaranteed correctness.Dr. AMIT KUMAR, Dept of CSE, JUET, Guna

Basic idea:
Knowledge encoded in some procedures
small programs that know how to do specific things,
how to proceed.
It is mainly operational knowledge that specifies “ what to do when”
There are many ways to represent procedural knowledge
LISP Coding
Does not have reasoning capability i.e. Inferential Adequacy
And Acquisitional Efficiency
Production rule:
Augment with the information on how to use them

Different procedural
representation• Frame Production Rule
• LISP Code

1. Important Attributes:
• Isa
• instance attributes.
Can be defined explicitly (frame representation) or implicitly (logic representation)
2. Relationships among attributes:
• inverses,: Can be represented as
 Team(pee-wee-reese, Brooklyn dodgers)
 Associate with pee-wee-reese : team = Brooklyn dodgers
Associate with Brooklyn dodgers: team-member = pee-wee-reese
• existence in a Isa hierarchy,
 Support inheritance property
 Helps to provide constrained information
Issues in Knowledge
Representation

• single-valued attributes,
• Support contradiction mechanism
• Overwrite the values
• techniques for reasoning about values.
• Information about the type of value eg. Height
• Constrained on the values(often stated in terms of other entities)
• Rules for computing the value when it is needed
Issues in Knowledge
Representation

3. Choosing the Granularity: High-level facts
may not be adequate for inference. Low-
level primitives may require a lot of storage.
Ex: “john spotted sue”
[representation:
spotted(agent(john),object(sue))]
Q1: “who spotted sue?”
Q2: “Did john see sue?”
Ans1: “john”.
Ans2: NO ANSWER!!!!
• Add detailed fact: spotted(x,y)-->saw(x,y) then Ans2: “Yes”.

• Advantage of using low level primitives:
1. The rules that are used to derive inference from the knowledge
need to be written in the terms of primitives only.
• Disadvantage of using low level primitives:
1. Require a lot of space to store, because some of the primitives are
just duplicates.
– Ex: john punched mary and mary punched jhon
1. the substantial amount of work is required to represent the
knowledge into low level primitives
2. Sometimes it is not clear what are the primitives
3. Ex domain: kinship
1. Primitives: mother father son daughter brother and sister
1. Problem how to represent Mary is sue’s cousin
– Mary = daughter(brother(mother(sue)))
– Mary = daughter(sister(mother(sue)))
– Mary = daughter(brother(father(sue)))
– Mary = daughter(brother(father(sue)))Dr. AMIT KUMAR, Dept of CSE, JUET, Guna

• 4.Representing Set of Objects:
• Must have certain ways to represent the set of objects because some
properties are true for set not for an indivisible
• Ex: English speakers can be found all over the world
• There are more sheep than people in Australia
• There are two ways to state the definition of a set and its elements
• Extensional definition
• List the member
• sun’s planets on which people live : {earth}
• Intensional definition
• Provide a rule that, when a particular object is evaluated, returns
true or false depending on whether the object is in set or not.
• sun’s planets on which people live
• {x: sun-planet(x) ^ human inhabited(x)}
• {x: sun-planet(x) ^ nth biggest(x,5)}

Advantage of intensional over Extensional
1. Can represent infinite set and the set whose all
elements are not known
1. Ex: set of prime numbers or kings of England
2. The intensional description can depend on time or
spatial location
1. President of India since 1945
1. Provide more crisp representation
Advantage of intensional over Extensional

5. finding the right structure as needed.:
How to select a proper structure to represent a fact:
Ex: Ram went to Grill last night. He ordered a grill sandwich, paid his
bill and left
Q.Did john eat dinner last night.
Approaches:
1. Index the structures directly by significant English words that can be
used to describe them
Ex: word “fly” can have multiple meanings:
1. “John flew to new york”
2. “John flew into a rage” [idiom]
3. “john flew a kite”
2. Consider the major concept as a pointer to all of the structures in which
it might be involved
Ex Steak : restaurant and supermarket
Bill: restaurant and shopping script
But how to deal with empty set
john road his bicycle to steak and ale last night.Dr. AMIT KUMAR, Dept of CSE, JUET, Guna

Knowledge representation

More Related Content

What's hot (20)

Similar to Knowledge representation (20)

More from Amit Kumar Rathi (20)

Recently uploaded (20)

Knowledge representation